Sei sulla pagina 1di 11

Fundamental Dimensions of Affective Responses to Product

Shapes

Kun-An Hsiao*, Lin-Lin Chen, Chi-Fu Wang, Huai-Te Tsang **

*Department of Industrial Design, Chang Gung University


Taiwan, kashau@mail.cgu.edu.tw
**Department of Industrial and Commercial Design, National Taiwan University of Science and
Technology
Taiwan, llchen@mail.ntust.edu.tw

Abstract: One of designers’ important tasks is to evoke specific affective responses via the manipulation of product
shapes. Nevertheless, how product shapes relate to these responses often remains a black box. In this research, we
selected three product categories—automobile, sofa, and kettle— representing large, medium, and small products, and
conducted three parallel studies to uncover fundamental dimensions of affective responses to product shapes. For each
product category, we collected a large number of product images and adjectives suitable for describing the affective
responses to these products. By using card sorting and clustering methods, we extracted representative products for
each category. We also generated a list of 28 adjectives covering all clusters of adjectives for the three product
categories. We then conducted three separate semantic differential surveys using the representative products and 28
adjectives. From the results deriving from factor analyses for the three product categories, we uncovered four
fundamental dimensions in the affective responses. The four dimensions are: the trend factor (T), the emotion factor
(E), the complexity factor (C), and the potency factor (P). The typical affective adjective pairs for four main
dimensions are: “contemporary- traditional” for factor T, “rational- emotional” for factor E, “simple- complex” for
factor C, and “strong- weak” for factor P. These fundamental dimensions underlying affective judgments of product
shapes could provide a common framework for studies on affective responses to product shapes.
Key words: product design, affective responses, factor analysis, clustering analysis, fundamental dimensions

1. Introduction
The physical shape of a product plays a critical role in its market success. As Bloch said: “A good design
attracts consumers to a product, communicates to them, and adds value to the product by increasing the quality of
the usage experiences associated with it” [1]. The manipulation of product shapes is therefore an important way
through which designers communicate messages and elicit responses from consumers. However, the relationship
between product shape and the affective responses often remains a black box. During the past decade, hundreds of
studies have been conducted to investigate this relationship, using Kansei Engineering and other methods, for a
wide range of products. Each of these studies usually devises its own set of scales for measuring emotional
responses for the specific product. The result is a set of tailor-made scales that shares few common scales with
other studies. This lack of common framework prevents these hundreds of Kansei studies from being compared in
parallel to discover structures, or to detect inconsistencies.
A common method used in many of these studies is the semantic differential method (SD), which was
developed by Osgood et al. to survey affective meanings [4]. A typical SD study evaluates the affective responses
elicited by product shapes using a set of bipolar adjective pairs. As a starting point for this research, we collected
17 studies that use semantic differential to extract the adjective pairs. We observed similar, but not directly
comparable, constructs in many of these studies, hinting the possible existence of a common framework in
consumers’ perception of product appearance. For example, factor 1 extracted in many of these studies using
factor analysis included adjectives such as avant-garde, modern, futuristic, stylish, contemporary, fashionable,
antique-looking, retro-looking and traditional [2,3,5,6]. All these adjectives seem to be related to the concept of
trendiness. The significance of these common adjectives were usually ignored, because these adjectives often
appear in a mixture with other adjectives. In this research, we attempt to uncover this common structure by
determining the fundamental dimensions underlying affective judgments of product shapes.

2. Representative Products and Adjectives


To investigate the affective responses evoked by product shapes, we selected three product categories,
automobile, sofa, and kettle that represent large, medium, and small products (with respect to human scale). For
each product category, we collected a large number of product images to cover, as much as possible, the range of
variations in product shapes. In total, 117 images of automobiles, 74 images of sofas and 64 images of kettles
were collected for the first stage.
Based on the collected product images, three surveys were conducted to elicit adjectives suitable for
describing affective responses to these products. Twenty subjects participated in the survey for each product
category. We extracted 719 affective adjectives for automobile, 510 affective adjectives for sofa, and 645 affective
adjectives for kettle from voice recordings of the three surveys. After eliminating similar and not suitable words,
we arrived at 100 affective adjectives for each product category.

19 Automobiles

20 Sofas

21 Kettles

Figure 1. Representative product images: 19 automobiles, 20 sofas and 21 kettles

The numbers of product images (117, 74, 64) and adjectives (100, 100, 100) were still too large to conduct
affective studies. We needed to extract representative products images and affective adjectives for each product
category. Clustering techniques are utilized for this purpose. Again, 20 subjects were assigned to the experiment
for each product category. The subjects were asked to sort cards of adjectives and cards of product images,
separately, into groups, each with similar affective meanings. The number of groups is limited to be at most 20.
After card sorting, subjects were asked to rank each card in the same group by the degree to which it represents
the group. Clustering method was then applied to analyze the card sorting data. For affective adjectives, we
obtained 13 groups for automobiles, 14 groups for sofa, and 11 groups for kettle. For product images, we obtained
19 groups for automobiles, 20 groups for sofas and 21 groups for kettles. Based on ranking data, we selected 19
automobiles, 20 sofas and 21 kettles to represent their groups (figure 1).
Using the ranking and clustering data for the affective adjectives, we extracted representative adjectives for
three product categories. Focusing on the total collection of representative adjectives for the three product
categories, we extracted 28 adjectives to describe affective reactions to all three products. By consulting related
studies and references, we composed 28 bipolar adjectives pairs, as shown in table 1.

Table 1. Twenty-eight bipolar affective adjectives


1 cute - not cute 15 rational - emotional
2 feminine - masculine 16 reliable - unreliable
3 futuristic - nostalgic 17 innovative - imitative
4 elegant - not elegant 18 heavy - light
5 mature - immature 19 varied - monotonous
6 dynamic - static 20 truthful - exaggerated
7 simple - complex 21 consistent - inconsistent
8 streamlined - rugged 22 aggressive - submissive
9 steady - unsteady 23 old - young
10 contemporary - traditional 24 gorgeous - plain
11 avant-garde - conservative 25 comfortable - uncomfortable
12 formal - casual 26 excited - calm
13 delicate - rough 27 strong - weak
14 dazzling - ordinary 28 soft - hard

3. Factor Analysis of Affective Responses


Using the representative product images and 28 bipolar adjectives for the three product categories obtained in
the previous stage, we conducted three separate semantic differential surveys. 30 subjects were assigned to
participate in the experiment for each product category. The questionnaires were in the same format for the three
surveys. Each sheet in a questionnaire shows one representative product image and 28 bipolar adjectives with
7-position scale. The positions were labeled slightly agree, agree, strongly agree, towards each extreme, and the
middle position was labeled “neutral”. Subjects marked the levels of affective response according to his or her
perception about the product image.
The results from the three surveys were analyzed using factor analysis for each of the three product categories.
Tables 2, 3 and 4 show the factor analysis results for kettle, sofa and automobile, respectively. The matrices of the
28 adjective pairs show 4 main factors and the loading for each adjective pair. The cumulative percentages are
high for each product category, indicating that the interpretative powers are high by using 4 factors (91.930% for
kettle, 89.004% for sofa and 91.066% for automobile). We observed that many adjective pairs appear in the same
factor for the three product categories. These findings confirm our observations about previous studies. Do there
exist common factors or in the affective responds when judging product shapes? We investigate this question in
the next section.
Table 2. Kettle: Matrix of 28 pairs adjectives by factor analysis
Factor 1 Factor 2 Factor 3 Factor 4
11 avant-garde - conservative 0.939 -0.239
17 innovative - imitative 0.932 -0.246
10 contemporary - traditional 0.931 0.157
14 dazzling - ordinary 0.928 0.194 -0.128
23 old - young -0.845 -0.461
3 futuristic - nostalgic 0.842 0.261 -0.139
19 varied - monotonous 0.763 0.198 -0.563
24 gorgeous - plain 0.724 0.214 -0.531 0.177
26 excited - calm 0.694 0.268 -0.596
13 delicate - rough 0.668 0.541 0.421
8 streamlined - rugged 0.135 0.979
28 soft - hard 0.953 -0.200 -0.114
2 feminine - masculine 0.291 0.926 -0.146
22 aggressive - submissive -0.921 -0.166 0.212
15 rational - emotional -0.346 -0.852 0.327
1 cute - not cute 0.477 0.841 0.156 0.114
25 comfortable - uncomfortable 0.825 0.442 0.207
4 elegant - not elegant 0.300 0.814 0.419 -0.141
6 dynamic - static 0.508 0.651 -0.494
21 consistent - inconsistent -0.109 0.897 0.188
16 reliable - unreliable -0.168 0.888 0.320
7 simple - complex -0.148 0.326 0.865 -0.149
5 mature - immature -0.193 0.342 0.813 -0.147
20 truthful - exaggerated -0.524 0.808 0.198
9 steady - unsteady -0.116 0.690 0.645
12 formal - casual -0.550 -0.291 0.621 -0.155
18 heavy - light 0.188 0.962
27 strong - weak 0.214 -0.417 0.132 0.840
Final Statistics
Eigenvalue 8.491 8.069 6.644 2.538
Pct of Var 30.323 28.817 23.727 9.063
Cum Pct 30.323 59.140 82.867 91.930

Table 3. Sofa: Matrix of 28 pairs adjectives by factor analysis


Factor 1 Factor 2 Factor 3 Factor 4
10 contemporary - traditional 0.954 -0.111 -0.161
17 innovative - imitative 0.947
11 avant-garde - conservative 0.942
20 truthful - exaggerated -0.882 -0.209 -0.311
3 futuristic - nostalgic 0.879 0.168 -0.150 -0.290
5 mature - immature -0.878 0.283 0.308
12 formal - casual -0.859 -0.294
23 old - young -0.853 -0.235 0.116 0.353
14 dazzling - ordinary 0.768 0.145 0.512
16 reliable - unreliable -0.728 -0.274 0.364
26 excited - calm 0.658 0.306 0.554 -0.253
21 consistent - inconsistent -0.577 -0.456 -0.557
28 soft - hard 0.931 0.195 -0.228
25 comfortable - uncomfortable -0.206 0.868 0.194 -0.195
8 streamlined - rugged 0.362 0.808 0.281
15 rational - emotional -0.469 -0.747 -0.398
1 cute - not cute 0.522 0.710 0.184 -0.306
6 dynamic - static 0.601 0.677 0.260
2 feminine - masculine 0.446 0.630 0.316 -0.462
24 gorgeous - plain 0.247 0.361 0.853
13 delicate - rough -0.318 0.123 0.785 -0.199
7 simple - complex -0.529 -0.721 0.105
4 elegant - not elegant -0.150 0.619 0.681 -0.157
19 varied - monotonous 0.482 0.445 0.578 -0.363
27 strong - weak -0.215 0.952
18 heavy - light -0.205 0.929
22 aggressive - submissive -0.389 -0.414 0.731
9 steady - unsteady -0.568 -0.323 -0.240 0.643
Final Statistics
Eigenvalue 10.531 6.034 4.478 3.878
Pct of Var 37.610 21.551 15.991 13.851
Cum Pct 37.610 59.161 75.153 89.004
Table 4. Automobile: Matrix of 28 pairs adjectives by factor analysis
Factor 1 Factor 2 Factor 3 Factor 4
6 dynamic - static 0.942 -0.161 0.168
11 avant-garde - conservative 0.913 0.301 -0.114
26 excited - calm 0.897 0.256 0.198
10 contemporary - traditional 0.893 0.249 -0.235
8 streamlined - rugged 0.840 0.159 0.323 -0.291
19 varied - monotonous 0.822 0.381 0.263
3 futuristic - nostalgic 0.818 -0.252 0.334 -0.174
23 old - young -0.772 0.263 -0.516
17 innovative - imitative 0.748 0.201 0.588
14 dazzling - ordinary 0.668 0.489 0.220 0.270
20 truthful - exaggerated -0.589 0.530 -0.422 -0.166
16 reliable - unreliable -0.211 0.904 -0.272 0.126
25 comfortable - uncomfortable 0.229 0.902 0.146
4 elegant - not elegant 0.240 0.900 0.220
13 delicate - rough 0.511 0.833
5 mature - immature -0.371 0.810 -0.409
9 steady - unsteady -0.460 0.786 -0.326 0.139
18 heavy - light -0.221 0.746 0.525
21 consistent - inconsistent -0.386 0.744 -0.417
27 strong - weak 0.162 0.691 -0.283 0.493
12 formal - casual -0.496 0.652 -0.511
24 gorgeous - plain 0.513 0.650 0.404
2 feminine - masculine 0.310 0.916 -0.165
1 cute - not cute 0.503 0.840
22 aggressive - submissive -0.279 0.343 -0.781 0.397
15 rational - emotional -0.609 0.187 -0.730
28 soft - hard 0.577 0.716 -0.249
7 simple - complex 0.147 0.288 -0.847
Final Statistics
Eigenvalue 9.973 7.754 5.493 2.279
Pct of Var 35.618 27.693 19.616 8.138
Cum Pct 35.618 63.311 82.927 91.066

4. Fundamental Dimensions
Table 5 shows the combination of 28 pairs adjective in 4 factors for three products categories. Because the
above experiments were conducted in parallel under similar conditions, we could compare the results derived
from factor analysis for the three product categories. Many adjective pairs obviously appear in the same factor for
the three products. Specifically, for factor 1, there exist 7 common adjective pairs: avant-garde – conservative,
innovative – imitative, contemporary – traditional, dazzling – ordinary, old – young, futuristic – nostalgic and
excited – calm. For factor 2 of kettle and sofa and factor 3 of automobile, there exist 4 common adjective pairs:
soft – hard, feminine – masculine, rational – emotional and cute - not cute. For factor 3 of kettle and sofa and
factor 4 of automobile, they have 1 common adjective pairs: simple – complex. For factor 2 of automobile and
factor 4 of kettle and sofa, there exist two common adjective pairs: heavy – light, strong – weak.
For factor 1, the seven common adjective pairs all relate to the evaluation of time, familiarity and trend.
Therefore, factor 1 is named the “trend factor”. This indicates that a typical reaction of a subject evaluating a
product’s shape is to compare the image with his or her experience, and determining whether the style is modern
or traditional.
For factor 2 of kettle and sofa, and factor 3 of automobile, there exist 4 common adjective pairs. These
adjectives all relate to the emotional character of a product. Therefore, this factor is named the “emotion factor”.
This factor has less number of common adjective pairs than factor 1. The same phenomenon shows up in factors 3
and 4, too. We find only one common adjective pair: simple – complex, in factor 3 of kettle, sofa and factor 4 of
automobile; and two common adjective pairs: heavy – light, strong – weak, in factor 4 of kettle, sofa and factor 2
of automobile.
Focusing on the common adjective pair for factor 3 ‘simple – complex’, this clearly relates to the shape
complexity when user perceives the product appearance. This factor is thus named the “complexity factor”. For
factor 4, the common adjective pairs: heavy – light and strong – weak relate to the psychological weight conveyed
by the product shapes. These results are in sync with the “potency factor” identified by Osgood in semantic
differential technique. Therefore, this factor is named the “potency factor”.
For different products, each of these factors may carry a slightly different meaning and may be realized via
slightly different shape characters. Even in factor 1, there exist adjectives for different product characters specific
to their product categories. Overall, we found that there seem to exist a few common perception factors in human
mind. We summarized these common factors as the “trend factor” (T), the “emotion factor” (E), the “complexity
factor” (C) and the “potency factor” (P).

Table 5. Comparison of 28 pairs adjective in 4 factors for three product categories


Factor 1 Factor 2 Factor 3 Factor 4
avant-garde - conservative streamlined - rugged consistent - inconsistent heavy - light
innovative - imitative soft - hard reliable - unreliable strong - weak
contemporary - traditional feminine - masculine simple - complex
Kettle dazzling - ordinary aggressive - submissive mature - immature
old - young rational - emotional truthful - exaggerated
futuristic - nostalgic cute - not cute steady - unsteady
varied - monotonous comfortable - uncomfortable formal - casual
gorgeous - plain elegant - not elegant
excited - calm dynamic - static
delicate - rough
contemporary - traditional soft - hard gorgeous - plain strong - weak
innovative - imitative comfortable - uncomfortable delicate - rough heavy - light
avant-garde - conservative streamlined - rugged simple - complex aggressive - submissive
truthful - exaggerated rational - emotional elegant - not elegant steady - unsteady
Sofa futuristic - nostalgic cute - not cute varied - monotonous
mature - immature dynamic - static
formal - casual feminine - masculine
old - young
dazzling - ordinary
reliable - unreliable
excited - calm
consistent - inconsistent
dynamic - static reliable - unreliable feminine - masculine simple - complex
avant-garde - conservative comfortable - uncomfortable cute - not cute
excited - calm elegant - not elegant aggressive - submissive
contemporary - traditional delicate - rough rational - emotional
streamlined - rugged mature - immature soft - hard
Automobile varied - monotonous steady - unsteady
futuristic - nostalgic heavy - light
old - young consistent - inconsistent
innovative - imitative strong - weak
dazzling - ordinary formal - casual
truthful - exaggerated gorgeous - plain
(Bold words with bottom line represent common adjective pair of three product categories)

To further understand the relationship between the common factors of affective responses and product shapes,
we matched the adjective pairs with product shapes for factor T, E, C and P for the three products. Figure 2 shows
the top five products for each of the three product categories as ranked by high loading value in factor analysis.
These products represent the product shapes for each group of affective adjectives and illustrate the meanings of
each factor.
Factor T Factor E Factor C Factor P
Trend factor Emotion factor Complexity factor Potency factor
avant-garde, conservative, soft, hard, simple complex heavy light
innovative imitative feminine, masculine, strong weak
Adjective

contemporary, traditional, emotional, rational,


dazzling ordinary cute not cute
young, old, nostalgic,
futuristic, calm
excited
Kettle
Sofa
Automobile

Figure 2. Adjective pairs and product images of factor A, B, C and D for three products

What kinds of shape features or shape manipulative methods might be the “potential influence factors”, that
help to create specific affective response? By observing the representational product images with related to their
adjectives in Figure 2, we have some tentative conclusions above the potential influence factors. For example, for
the Emotional factor, shapes with curve elements and smooth features usually project soft, feminine and emotional
images. On the other hand, shapes with the straight line elements, sharp corners and flat surfaces project hard,
masculine and rational images. Table 6 shows the potential influence factors for the four fundamental dimensions.
Comparing the contents of potential influence factors between four fundamental dimensions, it is relatively easy
to explain that how shape features could evoke specific affective responses for Emotion, Complexity and Potency
factors. But Trend factor is more difficult to explain from product shapes. We observed that the Trend factor
contains more abstract concepts (ex. symbolic elements, prototypical shapes, etc.), which go beyond simple
manipulation of shape elements and features.

Table 6. Potential influence factors for four fundamental dimensions.


Affective adjectives potential influence factors
(shape features and shape manipulation methods)
Trend factor avant-garde, innovative with symbolic elements, unlike prototypical shape
contemporary, dazzling, young,
futuristic, excited
conservative, imitative, traditional, functional elements, prototypical shape
ordinary, old, nostalgic, calm
Emotion factor soft, feminine, emotional, cute curve element, smooth features, organic surfaces
hard, masculine, rational, not cute straight line, sharp corner, flat surface
Complexity factor simple less elements, unified image, without decoration
complex more elements, compound image, with decoration
Potency factor heavy, strong more volume, stable elements
light, weak less volume, unstable elements
(Potential influence factors with bottom line represent abstract concept)

Factor E

Factor T

Figure 3. Perceptual space of factor E and factor T for three products

Figure 3 shows the combined perceptual space by factor analysis for factor T and factor E for three products.
Towards the righthand side are products with more traditional images, and towards the lefthand side products with
more contemporary images. Towards the top are products with more rational images, and towards the bottom
products with more emotional images.
From the distribution of products in figure 3, we observe that product shapes can evoke combinations of
affective responses corresponding to different factors at the same time. Figure 4 further demonstrates this
multi-faceted aspect of affective responses, illustrating how product shapes can present complex affective
expressions by shape manipulation. For example, products with contemporary image in factor T can present either
emotional or rational image in factor E. By the same token, products with emotional image in factor E can present
either contemporary or traditional image in factor T. But these affective expressions are integrated within each
product shape.

soft, feminine, emotional, cute hard, masculine, rational, not cute

avant-garde,
innovative
contemporary,
dazzling
young, future, excited

conservative,
imitative
traditional, ordinary
old, nostalgic,
calm
Figure 4. Cross comparison of product shapes and affective responses for factors T and E

To further examine the relationship between product shapes and affective responses on different shape
operations. We used two design methods, element recombination and shape interpolation, to create new product
shapes from existing products with same function and structure.
Figure 5 shows new shapes created using element recombination by exchanging handles of kettles A and B.
The two espresso moka kettles express different affective meanings: kettle A projects contemporary and emotional
images, and kettle B projects traditional and rational images. The recombined kettles in the middle appear to be
disharmonious and deliver confusing messages. Figure 6 shows another recombined shape created using the same
operation. Kettles C and D both exhibit ‘simple’ image in the Complexity factor. The recombined kettles appear to
be more reasonable and deliver a more coherent message. This example shows that if we choose product image
with similar affective characters that the recombined kettle might exhibit create more coherent images.

Kettle A Kettle B

Recombined Kettles
Figure 5. Element recombination with kettles A and B
Kettle C Kettle D

Recombined Kettles
Figure 6. Element recombination with kettles C and D

Figures 7 and 8 show the interpolation of two different kettle shapes by using morphing technique. These
morphing results appear to be more acceptable than the results obtained by element recombination. This is
because morphing maintains relationships among elements of a product and gradually changes its shape as a
whole from one affective expression to another. Comparing the results of figures 7 and 8, we find figure 8 to be
more realistic and is able to maintain contemporary image throughout the process.

avant-garde, conservative,
innovative imitative
contemporary, traditional, ordinary
dazzling old, nostalgic,
young, future, excited calm
+ +
soft, hard,
feminine, A100/B0 A75/B25 A50/B50 A25/B75 A0/B100 masculine,
emotional, rational,
cute Morphed Kettles (%A / %B) not cute

Figure 7. Morphing of kettles A and B

avant-garde, avant-garde,
innovative innovative
contemporary, contemporary,
dazzling dazzling
young, future, excited young, future, excited

A100/C0 A75/C25 A50/C50 A25/C75 A0/C100

Morphed Kettles (%A / %C)

Figure 8. Morphing of kettles A and C

5. Conclusion
The results of this study indicate the existence of a common perception mechanism in human’s mind when
making affective judgments of product shapes. From the results of factor analysis for three product categories, we
distilled four fundamental dimensions in these affective responses: the “trend factor” T, the “emotion factor” E,
the “complexity factor” C and the “potency factor” P. Typical affective adjective pairs for the four dimensions are:
“contemporary-traditional” for factor T, “rational-emotional” for factor E, “simple-complex” for factor C and
“strong – weak” for factor P. These fundamental factors/dimensions underlying affective judgments of product
shapes could serve as a common structure for further affective studies.
We also observed that the shape of a product can elicit multi-faceted affective responses. For example,
products projecting contemporary images can simultaneously project either soft or hard images. Therefore, when
analyzing these affective responses, the shape of a product needs to be examined as a whole and not as
independent elements.

Reference
1. Bloch, P. H. Seeking the ideal form: product design and consumer response. Journal of Marketing, 59, July,
16-29(1995).
2. Jindo, T., Hirasago, K. and Nagamachi, M. Development of a design support system for office chairs using
3-D graphics. International Journal of Industrial Ergonomics, 15, 49- 62(1995).
3. Kao, C. H. Exploring the relationship between the style and image and the goggles feature- from the style of
prototype. Journal of Design, 7(1), Summer (2002). (in Chinese)
4. Osgood, C.E., Suci, G. J., Tannenbaum, P.H., The Measurement of Meaning, University of Illinois
Press(1957).
5. Shang, H. H., Chuang, M. C., Chang, C. C. A semantic differential study of designers' and users' product form
perception. International Journal of Industrial Ergonomics, 25, 375-391(2000).
6. Toshimasa, Y., Toshiaki, U., Emi, H., Shigeo, H. Image database system based on readers Kansei character
[Electronic version]. 5th ADC Asian Design Conference, International Symposium on Design Science(2001).

Potrebbero piacerti anche